5

I am looking for a publicly available data set of physician notes that describe medical reasoning, ideally freetext indexed by the level of training of the author.

  1. MIMIC-III has physiologic data, and nursing and procedure notes. Procedure notes do not describe physician reasoning. They document the procedure but not why the procedure was done.
  2. IIB2B has a data set of about 1000 discharge notes. This is an acceptable starting point, but I was hoping for something larger.
| improve this question | |
  • MIMIC-III would appear to contain the kind of data that you are asking for - free text written by physicians - but you say that you are looking for a different type of note. To help find something that meets your needs, could you say more about why this resource is insufficient? – tomp Dec 10 '15 at 13:19
  • MIMIC has nursing noted and procedure notes. Procedure notes do not contain medical reasoning. A procedure note would say "I inserted a central venous catheter of these specifications into this location at this time with this anesthesia". A progress note/admit note/daily note would say "The patient needs a central venous catheter placed because X,Y, or Z." Also procedure notes need not be written by MDs (although one can filter based on author type) – mac389 Dec 10 '15 at 14:40
  • That makes sense, thanks for clarifying. The closest you would get to this in MIMIC would be the retrospective history of illness in discharge reports, which include comments like "patient developed hypotension that required levophed" and "temperature spiked to 100 so Flagyl was added for anaerobic coverage." – tomp Dec 10 '15 at 17:14
  • Discharge summaries might provide a good starting point. I hoped to be able to study the pattern of exposition of different authors (e.g. residents at different levels of training, the intern who recognized a high lactate but not sepsis). Discharge summaries would be a mishmash of different authors. – mac389 Dec 10 '15 at 20:52
  • 1
    In case it is useful, each note is associated with a Caregiver ID (CGID), which would give you some opportunity to explore patterns at an author level. The caregivers table associates CGIDs with different roles (e.g. MD, RN), though the role label is free text so would need some cleaning to be useful. – tomp Dec 10 '15 at 21:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.